It was right under my nose. I’d been optimizing my coding work with AI but completely neglecting everything else — emails, meetings, task management, research. The fix: connect Claude Desktop to all of it via MCP servers.
Standard connectors
The easy wins: Linear, Notion, Google Drive, Slack, Google Calendar, Google Tasks. Plug them in, Claude can read across all of them.
Custom MCP servers I built
The interesting part was building what didn’t exist yet:
- Gmail & bookmarks — Multi-account support, bookmark extraction via email
- Chrome tools — Bookmarks, history, screenshots, and HTML extraction for authenticated sites
- GitHub — With macOS Keychain and OAuth support
- Claude Code CLI history — Searchable local session parsing so Claude can reference past coding sessions
- Web processing — Jina AI for converting pages to clean markdown
- Keynote search — Making my archived presentations queryable
Each MCP took roughly 1.5 hours to build.
Voice changes everything
The real unlock was Wispr Flow for voice input. As a horrible typer, being able to talk to Claude across all these connected systems changed the interaction model completely.
Trust model: read-only first
Most of these MCPs operate read-only. I haven’t enabled autonomous actions yet — the value is already huge just from cross-referencing coding work, research, and presentations in one place.
What’s next
Exploring memory management with Mem0 and OpenMemory to help synthesize information and handle the context overload that comes from connecting this many sources.
Originally posted on LinkedIn